Bypass DRP if a redirect cycle is detected

components/data_reduction_proxy/core/common/data_reduction_proxy_headers.cc

 // Copyright 2014 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "components/data_reduction_proxy/core/common/data_reduction_proxy_headers.h"
 
 #include <stddef.h>
 #include <stdint.h>
 
 #include <string>
 #include <utility>
 #include <vector>
 
 #include "base/rand_util.h"
 #include "base/strings/string_number_conversions.h"
 #include "base/strings/string_split.h"
 #include "base/strings/string_util.h"
 #include "base/time/time.h"
 #include "components/data_reduction_proxy/core/common/data_reduction_proxy_event_creator.h"
 #include "components/data_reduction_proxy/core/common/data_reduction_proxy_params.h"
 #include "net/http/http_response_headers.h"
 #include "net/http/http_status_code.h"
+#include "net/url_request/url_request.h"
 
 using base::StringPiece;
 using base::TimeDelta;
 
 namespace {
 
 const char kChromeProxyHeader[] = "chrome-proxy";
 const char kChromeProxyECTHeader[] = "chrome-proxy-ect";
 const char kChromeProxyAcceptTransformHeader[] =
     "chrome-proxy-accept-transform";
@@ skipping 66 matching lines @@
                    const std::string& transform_type) {
   std::string header_value;
   if (!headers.GetNormalizedHeader(
           data_reduction_proxy::chrome_proxy_content_transform_header(),
           &header_value)) {
     return false;
   }
   return IsPreviewTypeInHeaderValue(header_value, transform_type);
 }
 
+// Returns true if there is a cycle in |url_chain|.
+bool HasURLRedirectCycle(const std::vector<GURL>& url_chain) {
+  if (url_chain.size() <= 1)
+    return false;
+
+  // If the last entry occurs more than once, then very likely there is a
+  // redirect cycle.
+  GURL last_entry = url_chain.back();
+  size_t last_entry_number_occurences = 0;
+
+  for (const auto& url : url_chain) {
+    if (url == last_entry) {
+      last_entry_number_occurences++;
+      if (last_entry_number_occurences >= 2)

[RyanSturm, 2017/03/28 18:03:19]

This looks confusing with the >=2 condition. Since you are O(n) of the chain
anyway, maybe you should just do something like:

vector<GURL> everything_but_the_back(url_chain.front(), url_chain.back());
for (const auto& url : everything_but_the_back) {
  if (url == last_entry)
    return true;
}
return false;


The most efficient option might be (it doesn't have to walk the whole chain or
copy the chain):

auto last_entry  = std::prev(url_chain.end());
for (auto url : url_chain) {
  if (url == last_entry)
    continue;
  if (*url == *last_entry)
    return true;
}
return false;

You could also re-write the above to use a for loop with the termination
condition of iter != std::prev(vector.end())


My syntax might be a little off.

[tbansal1, 2017/03/28 19:59:48]

Done.

+        return true;
+    }
+  }
+  return false;
+}
+
 }  // namespace
 
 namespace data_reduction_proxy {
 
 const char* chrome_proxy_header() {
   return kChromeProxyHeader;
 }
 
 const char* chrome_proxy_ect_header() {
   return kChromeProxyECTHeader;
@@ skipping 157 matching lines @@
         // the data reduction proxy's Via header.
         *has_intermediary = !(headers->EnumerateHeader(&iter, "via", &value));
       return true;
     }
   }
 
   return false;
 }
 
 DataReductionProxyBypassType GetDataReductionProxyBypassType(
-    const net::HttpResponseHeaders* headers,
+    const std::vector<GURL>& url_chain,
+    const net::HttpResponseHeaders& headers,
     DataReductionProxyInfo* data_reduction_proxy_info) {
   DCHECK(data_reduction_proxy_info);
-  if (ParseHeadersForBypassInfo(headers, data_reduction_proxy_info)) {
+
+  bool has_via_header = HasDataReductionProxyViaHeader(&headers, nullptr);

[RyanSturm, 2017/03/28 18:03:19]

Any chance you want to change HasDataReductionProxyViaHeader to accept a const
ref to headers?

[tbansal1, 2017/03/28 19:59:48]

I was planning to do it in a separate CL, but I have included those changes in
this CL itself. PTAL. Thanks.

+
+  if (has_via_header && HasURLRedirectCycle(url_chain)) {
+    data_reduction_proxy_info->bypass_all = true;
+    data_reduction_proxy_info->mark_proxies_as_bad = false;
+    data_reduction_proxy_info->bypass_duration = base::TimeDelta();
+    data_reduction_proxy_info->bypass_action = BYPASS_ACTION_TYPE_BLOCK_ONCE;
+    return BYPASS_EVENT_TYPE_URL_REDIRECT_CYCLE;
+  }
+
+  if (ParseHeadersForBypassInfo(&headers, data_reduction_proxy_info)) {
     // A chrome-proxy response header is only present in a 502. For proper
     // reporting, this check must come before the 5xx checks below.
     if (!data_reduction_proxy_info->mark_proxies_as_bad)
       return BYPASS_EVENT_TYPE_CURRENT;
 
     const TimeDelta& duration = data_reduction_proxy_info->bypass_duration;
     if (duration <= TimeDelta::FromSeconds(kShortBypassMaxSeconds))
       return BYPASS_EVENT_TYPE_SHORT;
     if (duration <= TimeDelta::FromSeconds(kMediumBypassMaxSeconds))
       return BYPASS_EVENT_TYPE_MEDIUM;
     return BYPASS_EVENT_TYPE_LONG;
   }
 
   // If a bypass is triggered by any of the following cases, then the data
   // reduction proxy should be bypassed for a random duration between 1 and 5
   // minutes.
   data_reduction_proxy_info->mark_proxies_as_bad = true;
   data_reduction_proxy_info->bypass_duration = GetDefaultBypassDuration();
 
   // Fall back if a 500, 502 or 503 is returned.
-  if (headers->response_code() == net::HTTP_INTERNAL_SERVER_ERROR)
+  if (headers.response_code() == net::HTTP_INTERNAL_SERVER_ERROR)
     return BYPASS_EVENT_TYPE_STATUS_500_HTTP_INTERNAL_SERVER_ERROR;
-  if (headers->response_code() == net::HTTP_BAD_GATEWAY)
+  if (headers.response_code() == net::HTTP_BAD_GATEWAY)
     return BYPASS_EVENT_TYPE_STATUS_502_HTTP_BAD_GATEWAY;
-  if (headers->response_code() == net::HTTP_SERVICE_UNAVAILABLE)
+  if (headers.response_code() == net::HTTP_SERVICE_UNAVAILABLE)
     return BYPASS_EVENT_TYPE_STATUS_503_HTTP_SERVICE_UNAVAILABLE;
   // TODO(kundaji): Bypass if Proxy-Authenticate header value cannot be
   // interpreted by data reduction proxy.
-  if (headers->response_code() == net::HTTP_PROXY_AUTHENTICATION_REQUIRED &&
-      !headers->HasHeader("Proxy-Authenticate")) {
+  if (headers.response_code() == net::HTTP_PROXY_AUTHENTICATION_REQUIRED &&
+      !headers.HasHeader("Proxy-Authenticate")) {
     return BYPASS_EVENT_TYPE_MALFORMED_407;
   }
-  if (!HasDataReductionProxyViaHeader(headers, NULL) &&
-      (headers->response_code() != net::HTTP_NOT_MODIFIED)) {
+  if (!has_via_header && (headers.response_code() != net::HTTP_NOT_MODIFIED)) {
     // A Via header might not be present in a 304. Since the goal of a 304
     // response is to minimize information transfer, a sender in general
     // should not generate representation metadata other than Cache-Control,
     // Content-Location, Date, ETag, Expires, and Vary.
 
     // The proxy Via header might also not be present in a 4xx response.
     // Separate this case from other responses that are missing the header.
-    if (headers->response_code() >= net::HTTP_BAD_REQUEST &&
-        headers->response_code() < net::HTTP_INTERNAL_SERVER_ERROR) {
+    if (headers.response_code() >= net::HTTP_BAD_REQUEST &&
+        headers.response_code() < net::HTTP_INTERNAL_SERVER_ERROR) {
       // At this point, any 4xx response that is missing the via header
       // indicates an issue that is scoped to only the current request, so only
       // bypass the data reduction proxy for the current request.
       data_reduction_proxy_info->bypass_all = true;
       data_reduction_proxy_info->mark_proxies_as_bad = false;
       data_reduction_proxy_info->bypass_duration = TimeDelta();
       return BYPASS_EVENT_TYPE_MISSING_VIA_HEADER_4XX;
     }
 
     // Missing the via header should not trigger bypass if the client is
@@ skipping 49 matching lines @@
   std::string value;
   size_t iter = 0;
   while (headers->EnumerateHeader(&iter, kChromeProxyHeader, &value)) {
     if (StartsWithActionPrefix(value, kChromeProxyActionFingerprintChromeProxy))
       continue;
     values->push_back(std::move(value));
   }
 }
 
 }  // namespace data_reduction_proxy